% fortune -ae paul murphy

An imaginary debate

Challenge: pick the right information architecture for a 10 to 15 FTE research foundation. The operating budget runs around two million per year, the mandate is to hand out about one hundred million per year in health care related research grants, the scope is nominally worldwide with preference given to people working in Alberta, western Canada, Canada, and the United States in that order. The executive director is a former provincial politician, the board a legislative committee.

The primary work process for the non administrative staff will consist of vetting, approving, tracking, and supporting health care related research. Thus these people will spend most of their time talking to researchers, most of whom are professors and graduate students, and be judged on a combination of paperwork and the results achieved by the rsearch they support.

There are two finalists for the CIO/IT role - and the director has called them in for a head to head discussion.

"George" has been nominated by a senior government IT manager who had been assigned as the IT liason for the director's legislature office during his earlier service as the minister responsible for Economic development and Technology. George assumes he'll be running a bunch of Windows servers and a small SAN in some kind of office data center, a wireless router or two, some HP printers, ten to fifteen desktops and the Vista/Office equiped Thinkpads to be issued each staffer.

When asked about software he mentions Microsoft Office with IIS, SQL-Server, and Active Directory on the backend. When asked about backup, he refers to the legislature's IT staff and suggests that any one of them can take over during emergencies or when he's on holidays.

The second candidate, "Paul", was brought in by the lady hired to become the group's office manager. She knows him for his work as a volunteer at the local Opera company where he'd set up and run their ticketing, accounting, and promotions system.

Asked to comment on what hardware and software the organization is likely to need he says that he can't answer that without making assumptions about how the organization will operate.

I've been assuming, he says, that you'll want to depend on researchers to evaluate both research proposals and work in progress - and that means the office will be the focal point for decision making, group meetings, and administrative support with most of the development officers working from their homes or the University facilities that go with their regular jobs.

If that's right, the office hardware and software decisions will be driven by the choice of funds accounting system - because everything else is pretty standard and stuff like Domino and OpenOffice run on just about anything.

For example, he says, Oracle's stuff may be a good choice because the government has a site license for it - making the group's choice free to the taxpayer - and the auditor general won't hassle anyone over the choice because it's already familiar.

So, he says, assuming that this is how you organize the business and that the software you choose for funds accounting works on it, the right solution is probably his favorite: two Solaris servers with the administration package and stuff like Apache, Domino, and OpenOffice, one in the group's offices and one on another ISP somewhere else; Sun Rays in the offices, meeting rooms, and available for home use to development officers; and playphones like Apple's iPhone for on the go use by everyone.

That, he says, provides maximum flexibility at minimal continuing cost - an hour or two a week in support and he can name several companies in town who can take contracts to provide the service.

George responds by saying that Paul's ideas are unproven, Unix is expensive, the software is all either archaic or unsupported, and nobody wants to go back to dumb terminals - instead, he says, staff will absolutely need and demand notebooks for out of office use and many professionals simply won't consider jobs that don't come with personal notebooks. The government, he adds, has site licenses for virtually all Microsoft software, so the group will face no incremental licensing cost - except for billbacks on license cost sharing.

Paul's response is that yes, his system's initial cost will be higher but includes cell phones, home office support for development officers; real protection against data losses; enough redundancy to continue essentially uninterupted while the office building burns down; and, that his stuff goes in a closet instead of a data center while producing better desktop performance with less heat and noise than PCs.

More, he says, George's system cost needs to include George along with George's office and the cost of working around the weekly or monthly security, software, or continuity crisises common to Wintel installations. "Did you ever" he asks the director, "have a week go by as a minister without having IT come in to fix something? - or a month without an IT crisis or new spending request?"

It's true, he says, that the iPhone approach is still largely unproven - but so were rubber tires and transisters in their time. What users need, he argues, is just about here in the first generation product: email, notes, web access to both the internet and centralized data, pretty good communications security, and easy desktop syncronization - just about the only limitations on what it can do right now, he says, are syncronization problems with Microsoft's email and other servers (which don't affect Domino or Apache) and the fact that upper case, lower case, and numeric keypad switching is so messy that people tend to use over simplified passwords. That'll be fixed soon, he says, and in a year or two they'll probably get automatic voice to text conversion, better cryptology, plugin custom application cards - possibly even display projection - although that might be four or five years out.

George responds that none of that's proven: plugin what everybody else has now and get on with the job, he says. Otherwise there'll be no support from the government wide IT people - and Public Works, he says, will probably reject a Unix recommendation anyway.

Paul agrees that his approach involves a risk of failure - but argues that adopting Windows gives up the opportunity for IT success and accepts known frustrations, security failures, and costs in exchange for the near certainty that you don't get fired for buying what everybody else does. Go with George, he tells the director, and no one will question your judgement; pick something different and you get the opportunity to remove systems as a daily irritant, but you'll wear the decision if things go wrong.

What it comes down to, he says, is that some risks are worth taking - and the opportunity here to marry well proven and highly reliable technology - Solaris, Sun Ray - to new and exciting technologies - the iPhone- is one of those risks worth taking. The fundamental organizational design implication, he says, is that you can have more people from the sharp end doing more of the heavy lifting at lower total cost - and that, he says, is really the bottom line: if systems can help your decision makers be more effective, than that's the gain that makes the risk worth taking.

George says every other provincially funded agency uses Windows and it would be insane to diverge from what works for everyone else.

The director - is a politician and asks for your advice...

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.